Cancer biomarkers are at the leading edge of Precision Medicine, and offer both tremendous opportunities and challenges. In particular, biomarker development to detect recurrence in cancer survivors is blossoming, as surveillance testing with serial biomarker measurements offers an opportunity to detect recurrence at a point when treatment may be curative. However, frequent biomarker testing may cause more harm than benefit for low-risk individuals, due to the costs and complications of unnecessary testing and increased likelihood of false positives leading to unnecessary treatment. Unfortunately, tailoring surveillance to individual patients is a complex decision-making problem that requires understanding the heterogeneity in biomarker measurements across patients and across time within patients. As a result, surveillance testing guidelines using one-size-fits- all strategies continue to be common in most cancers, despite their uncertain clinical utility. The overarching goal of the proposed research is to develop a decision-making framework to identify optimal surveillance strategies among cancer survivors.
The specific aims are:
Aim 1. Develop and evaluate a sequential decision- making framework by merging statistical methods for prediction modeling with economics concepts for value of information (VOI) analysis to guide individualized decisions about testing and treatment for recurrence using serial biomarker testing, with the goal of optimizing long-term patient outcomes.
This aim will build on preliminary work and develop a dynamic decision-making algorithm that uses accumulated information at a given time to update predictions and guide decisions. The broad applicability of the framework will be demonstrated by considering three distinct cancer surveillance settings: colorectal cancer (CRC), prostate cancer (PrCA), and chronic myeloid leukemia (CML), which capture a range of decision-making problems in cancer surveillance.
Aim 2. Apply this framework to existing electronic health record (EHR) and cohort study data to identify a risk-adaptive surveillance strategy for detecting CRC recurrence that targets high-risk patients for frequent follow-up and treatment, and recommends less frequent follow-up for low-risk patients.
Aim 3. Assess the comparative effectiveness of the proposed risk-adaptive surveillance strategy versus guideline- based surveillance in CRC.
Aim 4. Use existing data to evaluate the generalizability of the framework by addressing the optimal frequency of follow-up among (a) low-risk men with recurrent PrCA, for whom treatment may be safely delayed for a prolonged period, and (b) long-term survivors of CML, who achieve long-term remission but currently continue to be monitored frequently. We address a significant problem in cancer survivorship care using approaches to help resolve the uncertainty that clinicians and patients face when confronted with using new and evolving biomarker technologies to monitor for recurrence after patients have survived their primary cancer.
This research addresses a significant problem in cancer survivorship care by using novel approaches to help resolve the uncertainty that clinicians and patients face when confronted with using new and evolving biomarker technologies to monitor for recurrence after patients have survived their primary cancer. Our holistic approach of applying the decision-making framework to three wide ranging cancer applications will inform both clinical decision-making regarding optimal patient follow-up and also future reimbursement policies.
